(originally posted at TechPresident)

The study Show Us the Stimulus (July 2009, Good Jobs First) is one of the most comprehensive and systematic assessments of US state “recovery” websites. The authors of the report analyze the effectiveness and transparency of state websites in providing information on the different categories of stimulus spending, the allocation of funds across different areas of the state, and individual projects carried out by private contractors and their respective impact on employment levels.

The study shows that, while some websites achieve satisfactory levels of transparency, others are largely failing to provide online transparency with regard to the use of crisis response funds. Such variance among the websites per se is not particularly surprising. But why do some states perform better than others? Are there any factors that can help to explain these differences?

In a rather exploratory approach, I have looked for a correlation between the transparency scores achieved by each state in the study and factors that could be considered as likely to influence the provision of online transparency.

For instance, one possible hypothesis is that the variance in the provision of online transparency of recovery spending across states is related to the amount of funds received by each state. That is, states that receive more funds would be prompted – or pressured – to make more of an effort to disclose recovery information than those receiving fewer funds. Nevertheless, no statistical relationship can be found between the relative (or absolute) amount of stimulus funds received by each state and the level of transparency of recovery websites.

In a similar vein, it could be hypothesized that a state affected by higher unemployment levels would be more inclined to communicate the recovery efforts being made to its population. However, as illustrated in the figure below, no correlation can be found between the level of unemployment and the transparency offered by the state recovery websites.

One could reasonably expect that a state with a more developed e-government structure would be more likely to convey more online transparency with regard to recovery spending. However, state governments that perform better in terms of the general delivery of online information and services are not necessarily providing the best online means for monitoring recovery funds. The same absence of correlation is identified when considering other factors such as the level of Internet access in a state, tax systems, per capita income, population and size of economy (GSP).

If most of the factors I have looked at have proven to be uncorrelated to the transparency of recovery websites, partisanship has shown itself to be correlated to transparency. More precisely, it is possible to identify a positive and significant correlation between the transparency of recovery websites and the percentage of seats held by Democrats in the lower house of state legislatures. In other words, the more seats held by Democrats in a state legislature, the more likely the state recovery website is to be transparent.

The scatter plot below illustrates this correlation. The horizontal line inside the diagram indicates the average (mean) transparency score of each state, with the points above the horizontal line representing the states scoring above the average and vice-versa. The vertical line divides those states with Republican (left side) and Democrat (right side) majorities in each lower house.

As the diagram illustrates, with the exception of only two cases (Virginia and South Carolina), all of the recovery websites scoring above the average in terms of information provided belong to states with a democrat majority in the lower house. This relationship holds true and statistically significant even when controlling for other factors such as the size of the state economy, e-government readiness and levels of Internet access. The results indicate that there is a less than 0.3% chance that this relationship is spurious.

Nevertheless, such a relationship is by no means perfect. For instance, it is possible to identify a large number of states with a Democratic majority in the lower house that present low levels of transparency (bottom right corner of the diagram). In this case, two outliers stand out: Illinois and Hawaii. With regard to the state of Illinois, despite the high percentage of democrat legislators in both houses, its recovery website scores no points whatsoever.

Even though it is not my intent to find a definitive explanation for outlying cases, it is important to remember that the evaluation of the websites was conducted just a few months after the political turmoil that led to Blagojevich’s removal from office (July 2009). With regard to Hawaii’s case, the only notable factor from a party perspective is that despite having a majority of Democrats in the state legislature, the state’s governor is a Republican. However, even if we choose to disregard these two cases, the pattern of the relationship between partisanship and transparency exposes other outliers, suggesting the obvious: a high level of Democratic control in a legislature is not the only factor affecting transparency levels of state recovery websites.

To conclude, the causation mechanism linking partisanship and transparency deserves further study. For instance, if recovery efforts were to be launched by a Republican President, would we see a different picture, with states of republican legislative majorities offering more transparent recovery websites? Or is transparency more of a democrat virtue?

(Note: These findings are, of course, incipient results of a broader study looking at factors that might influence transparency policies at the macro and micro-levels. Suggestions of other factors I should be looking at are very welcome.)